]> Cypherpunks repositories - gostls13.git/log
gostls13.git
3 years agoall: update vendored golang.org/x/tools for Go 1.18 release
Dmitri Shuralyov [Tue, 9 Nov 2021 18:54:57 +0000 (13:54 -0500)]
all: update vendored golang.org/x/tools for Go 1.18 release

The Go 1.18 code freeze has recently started. This is a time to update
all golang.org/x/... module versions that contribute packages to the
std and cmd modules in the standard library to latest master versions.

This CL updates only the tools module, keeping mod unchanged because
its lastest commit isn't ready to be vendored yet.

For #36905.
Updates #49350.

Change-Id: Ib39713d28a55fc9ec79058aab9919eba912def5f
Reviewed-on: https://go-review.googlesource.com/c/go/+/361094
Trust: Dmitri Shuralyov <dmitshur@golang.org>
Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agoall: update vendored golang.org/x/crypto for Go 1.18 release
Dmitri Shuralyov [Tue, 9 Nov 2021 18:49:45 +0000 (13:49 -0500)]
all: update vendored golang.org/x/crypto for Go 1.18 release

The Go 1.18 code freeze has recently started. This is a time to update
all golang.org/x/... module versions that contribute packages to the
std and cmd modules in the standard library to latest master versions.

This CL updates only the crypto module, as well as the TestDependencies
policy to accommodate the dependency order change done in CL 345649.
The next CL will update further modules.

For #36905.

Change-Id: If93160d8d72ca86e9995aaf5bdfc3d2c20b4695d
Reviewed-on: https://go-review.googlesource.com/c/go/+/362736
Trust: Dmitri Shuralyov <dmitshur@golang.org>
Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
3 years agoall: update vendored golang.org/x/{net,text} for Go 1.18 release
Dmitri Shuralyov [Tue, 9 Nov 2021 18:31:45 +0000 (13:31 -0500)]
all: update vendored golang.org/x/{net,text} for Go 1.18 release

The Go 1.18 code freeze has recently started. This is a time to update
all golang.org/x/... module versions that contribute packages to the
std and cmd modules in the standard library to latest master versions.

This CL updates only the net, text modules.
The next CL will update further ones.

For #36905.

Change-Id: I9a5ac3cca22da961cfd09f3202e01e1187d42bdd
Reviewed-on: https://go-review.googlesource.com/c/go/+/362735
Trust: Dmitri Shuralyov <dmitshur@golang.org>
Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
3 years agoall: update vendored golang.org/x/{arch,sys,term} for Go 1.18 release
Dmitri Shuralyov [Tue, 9 Nov 2021 18:31:45 +0000 (13:31 -0500)]
all: update vendored golang.org/x/{arch,sys,term} for Go 1.18 release

The Go 1.18 code freeze has recently started. This is a time to update
all golang.org/x/... module versions that contribute packages to the
std and cmd modules in the standard library to latest master versions.

This CL updates only the lower-level modules arch, sys, term for better
bisection. The next CL will update further ones.

For #36905.

Change-Id: I455428c051ec49b446b8b558a6f579cd9be4d796
Reviewed-on: https://go-review.googlesource.com/c/go/+/362734
Trust: Dmitri Shuralyov <dmitshur@golang.org>
Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agoos: clarify that File.{Read,Write} use the buffer
Ian Lance Taylor [Tue, 9 Nov 2021 02:30:30 +0000 (18:30 -0800)]
os: clarify that File.{Read,Write} use the buffer

Fixes #49470

Change-Id: I81fd4b0e2eef1d8d430b5d1d10c4f824e803a75c
Reviewed-on: https://go-review.googlesource.com/c/go/+/362335
Trust: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
3 years agobufio: document that NewWriter can return its argument
Ian Lance Taylor [Mon, 8 Nov 2021 21:57:50 +0000 (13:57 -0800)]
bufio: document that NewWriter can return its argument

Fixes #49446

Change-Id: Ib0b53a7dd5d567a2dd0bdf29f53d276587b60afb
Reviewed-on: https://go-review.googlesource.com/c/go/+/361921
Trust: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Tobias Klauser <tobias.klauser@gmail.com>
3 years agocmd/compile: NewSelectorExpr use n.Pos instead of base.Pos
Chaoqun Han [Mon, 8 Nov 2021 14:58:51 +0000 (22:58 +0800)]
cmd/compile: NewSelectorExpr use n.Pos instead of base.Pos

Fixes #49436

Change-Id: I4c8851e7aaee631d5eb22f2ef0aea5a25e936d87
Reviewed-on: https://go-review.googlesource.com/c/go/+/361917
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Trust: David Chase <drchase@google.com>

3 years agogo/types: make object test an external test
Robert Findley [Tue, 9 Nov 2021 16:19:20 +0000 (11:19 -0500)]
go/types: make object test an external test

This is a port of CL 361409 to go/types.

Change-Id: I17ccf8a5b4ba715fd8a87ea2c1811700fb1157e3
Reviewed-on: https://go-review.googlesource.com/c/go/+/362538
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agogo/types: check non-generic conversions first
Robert Findley [Tue, 9 Nov 2021 16:14:37 +0000 (11:14 -0500)]
go/types: check non-generic conversions first

This is a clean port of CL 361269 to go/types.

Change-Id: I2caaf08eabdf1707ae83ec1e628fd26f21b2b8e5
Reviewed-on: https://go-review.googlesource.com/c/go/+/362616
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agogo/types: minor cleanups in predicates.go
Robert Findley [Tue, 9 Nov 2021 16:09:13 +0000 (11:09 -0500)]
go/types: minor cleanups in predicates.go

This is a clean port of CL 360956 to go/types.

Change-Id: Iac437e72bb760e7e90236a86e7473d6a440df081
Reviewed-on: https://go-review.googlesource.com/c/go/+/362615
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agocmd/link: fix GCC startfiles names on AIX
Clément Chigot [Tue, 9 Nov 2021 09:01:05 +0000 (10:01 +0100)]
cmd/link: fix GCC startfiles names on AIX

Since GCC version 11, the 64-bit version of GCC starting files are
now suffixed by "_64" instead of being stored without suffix under
"ppc64" multilib directory.

Change-Id: Ibe53521ed24d36e5f6282e3574849b9ae11a1e9a
Reviewed-on: https://go-review.googlesource.com/c/go/+/362594
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Ian Lance Taylor <iant@golang.org>

3 years agonet/http: add MaxBytesHandler
Carl Johnson [Tue, 31 Aug 2021 20:35:35 +0000 (20:35 +0000)]
net/http: add MaxBytesHandler

Fixes #39567

Change-Id: I226089b678a6a13d7ce69f360a23fc5bd297d550
GitHub-Last-Rev: 6435fd5881fc70a276d04df5a60440e365924b49
GitHub-Pull-Request: golang/go#48104
Reviewed-on: https://go-review.googlesource.com/c/go/+/346569
Trust: Damien Neil <dneil@google.com>
Trust: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Damien Neil <dneil@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Damien Neil <dneil@google.com>
3 years agocmd/go: add //go:embed all:pattern
Russ Cox [Thu, 28 Oct 2021 17:22:07 +0000 (13:22 -0400)]
cmd/go: add //go:embed all:pattern

When //go:embed d matches directory d, it embeds the directory
tree rooted at d, but it excludes files beginning with . and _,
as well as files having problematic names that will not be packaged
into modules (names such as .git and com1).

After long discussions on #42328 and #43854, we decided to keep
the behavior of excluding . and _ files by default, but to allow the pattern
prefix 'all:' to override this default. This CL implements that change.

Note that paths like .git and com1 are still excluded, as they must be,
since they will never be packed into a module.

Fixes #43854.

Change-Id: I4f3731e14ecffd4b691fda3a0890b460027fe209
Reviewed-on: https://go-review.googlesource.com/c/go/+/359413
Trust: Russ Cox <rsc@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jonathan Amsterdam <jba@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agoruntime: add upper half and carry bit with zero
Michael Pratt [Tue, 9 Nov 2021 16:35:52 +0000 (11:35 -0500)]
runtime: add upper half and carry bit with zero

https://golang.org/cl/246763 accidentally changed this from upper + 0 +
carry to upper + old vdsoSP + carry.

The old value of vdsoPC is usually zero, so this typically works.
However, the reentrant case will have a non-zero value, resulting in a
bogus returned time.

Fixes #49481

Change-Id: I0110b84277bf911804cb0ff8097aebf1b7eb100a
Reviewed-on: https://go-review.googlesource.com/c/go/+/362674
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Michael Pratt <mpratt@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agoruntime/pprof: include labels in profile dump
Michael Pratt [Tue, 9 Nov 2021 15:59:32 +0000 (10:59 -0500)]
runtime/pprof: include labels in profile dump

For tests of pprof label support having the sample labels in the output
is needed for effective debugging.

For #48577

Change-Id: Ic7c5bc90cb33e8fb477f7db62d9b56a7a9d6ffa8
Reviewed-on: https://go-review.googlesource.com/c/go/+/362614
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Michael Pratt <mpratt@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agocmd/compile/internal/types2: roll-forward removal of asX converters
Robert Findley [Mon, 8 Nov 2021 18:08:59 +0000 (18:08 +0000)]
cmd/compile/internal/types2: roll-forward removal of asX converters

This CL reverts CL 361964, rolling forward the original CL 362254 with a
fix for re-entrant expansion via type hashing (compare patchsets 1 and
2).

Change-Id: I62869e50e919f42eb8d6fef5b0d7a5ec8960bd84
Reviewed-on: https://go-review.googlesource.com/c/go/+/362118
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/compile: disable doubleCheck and remove unused checkDictionary
Dan Scales [Tue, 9 Nov 2021 02:10:30 +0000 (18:10 -0800)]
cmd/compile: disable doubleCheck and remove unused checkDictionary

In stencil.go, change doubleCheck to false, which will turn off some
double-checking code which isn't needed generally, now that we have lots
of tests of end-to-end runs.

Also, removed checkDictionary() which is unused and is unlikely to be
useful in the future.

Change-Id: I4e5acceab80f4904b174422bae21ca82cf04f943
Reviewed-on: https://go-review.googlesource.com/c/go/+/361923
Trust: Dan Scales <danscales@google.com>
Run-TryBot: Dan Scales <danscales@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agogo/types: rename is_X predicates back to isX (step 2 of 2)
Robert Findley [Tue, 9 Nov 2021 15:20:50 +0000 (10:20 -0500)]
go/types: rename is_X predicates back to isX (step 2 of 2)

This is a port of CL 361134 to go/types.

Change-Id: Ibac4365a85561b32a90b0118d48aa9302f227b2e
Reviewed-on: https://go-review.googlesource.com/c/go/+/362554
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agogo/types: rename isX predicates to allX, add simple is_X (step 1 of 2)
Robert Findley [Mon, 8 Nov 2021 18:09:41 +0000 (13:09 -0500)]
go/types: rename isX predicates to allX, add simple is_X (step 1 of 2)

This is a port of CL 360955 to go/types. Note that go/types and types2
differ in handling of untyped nil within both Checker.shift and
Checker.implicitTypeAndValue.

A missing comment was added to Checker.indexExpr.

Change-Id: Ia9149ff9c0af68213c579090902ab7989828ddd2
Reviewed-on: https://go-review.googlesource.com/c/go/+/362534
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agotesting: remove package from fuzz crasher message
Katie Hockman [Mon, 8 Nov 2021 17:13:01 +0000 (12:13 -0500)]
testing: remove package from fuzz crasher message

Fixes #48149

Change-Id: Iaf91d2c54fda809c7da90cdfb6d1d075f474c69b
Reviewed-on: https://go-review.googlesource.com/c/go/+/362116
Trust: Katie Hockman <katie@golang.org>
Run-TryBot: Katie Hockman <katie@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Julie Qiu <julie@golang.org>
3 years agocmd/go: fix mod_get_direct
Cuong Manh Le [Tue, 9 Nov 2021 03:37:43 +0000 (10:37 +0700)]
cmd/go: fix mod_get_direct

https://github.com/googleapis/google-cloud-go has changed the default
branch from master to main, causing mod_get_direct failed on longtest.

Change-Id: I8fe0356b2ff532d1fdedbcb1e1832d7335babaa0
Reviewed-on: https://go-review.googlesource.com/c/go/+/361965
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agoruntime: revert recent Windows crashdump changes
Jason A. Donenfeld [Tue, 9 Nov 2021 00:23:02 +0000 (01:23 +0100)]
runtime: revert recent Windows crashdump changes

Recent changes to runtime enabled crashdumps, which under some
circumstances apparently might result in memory being uploaded to
Microsoft. A change like this should go through the proper proposals
process where we can discuss how to gate it and what all of its
implications are.

This reverts CL 307372 and its cleanup CL 360617.

Change-Id: If2e74015899d746831da40546c82eacacdf739e1
Reviewed-on: https://go-review.googlesource.com/c/go/+/362454
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Run-TryBot: Jason A. Donenfeld <Jason@zx2c4.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agocmd/compile: remove unneeded "==" method in pre-defined "comparable" interface
Cuong Manh Le [Mon, 8 Nov 2021 08:51:05 +0000 (15:51 +0700)]
cmd/compile: remove unneeded "==" method in pre-defined "comparable" interface

Fixes #49421

Change-Id: Iecf3952346ecd278198c1000014a321e230f7fa7
Reviewed-on: https://go-review.googlesource.com/c/go/+/361962
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Trust: Dan Scales <danscales@google.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
3 years agocmd/compile: fix irgen mis-handling of ... argument when creating closure
Cuong Manh Le [Mon, 8 Nov 2021 06:08:58 +0000 (13:08 +0700)]
cmd/compile: fix irgen mis-handling of ... argument when creating closure

When bulding formal arguments of newly created closure, irgen forgets to
set "..." field attribute, causing type mismatched between the closure
function and the ONAME node represents that closure function.

Fixes #49432

Change-Id: Ieddaa64980cdd3d8cea236a5a9de0204ee21ee39
Reviewed-on: https://go-review.googlesource.com/c/go/+/361961
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Trust: Dan Scales <danscales@google.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
3 years agoRevert "cmd/go: add workspace pruning mode"
Bryan C. Mills [Mon, 8 Nov 2021 21:45:58 +0000 (21:45 +0000)]
Revert "cmd/go: add workspace pruning mode"

This reverts CL 357169.

Reason for revert: appears to be failing on longtest SlowBot.¹

¹https://storage.googleapis.com/go-build-log/a97c855b/linux-amd64-longtest_7c9857d4.log

Change-Id: I3b94395671db78ed5fb2fb1019e7199e4ffbd272
Reviewed-on: https://go-review.googlesource.com/c/go/+/362249
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Michael Matloob <matloob@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agotest: add regress test for reported non-monomorphizable example
Matthew Dempsky [Mon, 8 Nov 2021 21:13:30 +0000 (13:13 -0800)]
test: add regress test for reported non-monomorphizable example

Fixes #48711.

Change-Id: I2ed1ef5267343d4b9f91da0618905098d178db90
Reviewed-on: https://go-review.googlesource.com/c/go/+/362394
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agocmd/go: rename go mod initwork and editwork to go work init and edit
Michael Matloob [Thu, 21 Oct 2021 18:26:55 +0000 (14:26 -0400)]
cmd/go: rename go mod initwork and editwork to go work init and edit

This brings the workspace related commands into their own namespace.

Fixes #48256

Change-Id: I9d5e3d4c45798913d742c532c1571006e6f9cc57
Reviewed-on: https://go-review.googlesource.com/c/go/+/357611
Trust: Michael Matloob <matloob@golang.org>
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agocmd/go: add workspace pruning mode
Michael Matloob [Tue, 19 Oct 2021 18:05:29 +0000 (14:05 -0400)]
cmd/go: add workspace pruning mode

This change corrects a bug in the handling of module loading of
workspaces. Namely, there is an assumption by the module pruning code
that if a root module is selected then the packages of that module can
be resolved without loading the whole module graph. This is not true
in workspace mode because two workspace modules can require different
versions of a dependency. Worse, one workspace module can directly
require a depencency that is transitively required by another
workspace module, changing the version of that module loaded in the
fully expanded graph.

To correct this, a new 'workspace' pruning mode is added where the
roots are the workspace modules themselves, satisfying the assumption
made by the module pruning logic.

The rest of this change accounts for the new pruning mode where it's
used and correctly sets the requirements in this pruning mode.

Change-Id: I5d4d9877e492e196681f6ee9f8f18a08b4e95c61
Reviewed-on: https://go-review.googlesource.com/c/go/+/357169
Trust: Michael Matloob <matloob@golang.org>
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agonet/http: deflake TestTimeoutHandlerContextCanceled
Damien Neil [Mon, 8 Nov 2021 20:45:41 +0000 (12:45 -0800)]
net/http: deflake TestTimeoutHandlerContextCanceled

Fixes #49448

Change-Id: Ie2acff7dedbca9bd1cc0b1b3dd0a01573c7befee
Reviewed-on: https://go-review.googlesource.com/c/go/+/361920
Trust: Damien Neil <dneil@google.com>
Run-TryBot: Damien Neil <dneil@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agocmd/compile,cmd/go: fix long test builders
Cuong Manh Le [Mon, 8 Nov 2021 17:05:58 +0000 (00:05 +0700)]
cmd/compile,cmd/go: fix long test builders

CL 361411 improved error message for go version requirement, but forgot
to update the test in cmd/go to match new error message. That causes
longtest builders failed.

This CL changes mod_vendor_goversion.txt to match compiler error, and
limit fixedbugs/issue49368.go to run with -G=3 only.

Updates #49368

Change-Id: I125fe0a8c2a1595066d39c03e97819e7a1274e0a
Reviewed-on: https://go-review.googlesource.com/c/go/+/361963
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agocmd/compile: Revert "cmd/compile/internal/types2: remove most asX converters (cleanup)"
Cuong Manh Le [Mon, 8 Nov 2021 17:34:18 +0000 (00:34 +0700)]
cmd/compile: Revert "cmd/compile/internal/types2: remove most asX converters (cleanup)"

This reverts commit 759eaa22adb0ab883959e4a36c19f2dfe77b5895.

Reason to revert: break unified IR builder

Though the unified IR is not for go1.18, it's the only user of types2
lazy resolution APIs at this moment. And it consistently failed after
CL 362254 is the sign that the change was wrong somehow.

Change-Id: I6bfc3192904fe2129fd3c165f0df8761e8eb441c
Reviewed-on: https://go-review.googlesource.com/c/go/+/361964
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocrypto/x509: remove ios build tag restriction
Roland Shoemaker [Mon, 8 Nov 2021 16:05:24 +0000 (08:05 -0800)]
crypto/x509: remove ios build tag restriction

Fixes #49435

Change-Id: I77ce12f447e727e7dc3b23de947357c27a268bd2
Reviewed-on: https://go-review.googlesource.com/c/go/+/362294
Trust: Roland Shoemaker <roland@golang.org>
Run-TryBot: Roland Shoemaker <roland@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agointernal/fmtsort: order channels in test in memory address order
Keith Randall [Mon, 8 Nov 2021 05:29:30 +0000 (21:29 -0800)]
internal/fmtsort: order channels in test in memory address order

Kind of a kludge, but it makes the test work reliably.

Fixes #49431

Change-Id: Ic2a075ba02f80ea7efcc1b3f0f5a43649e87c0d8
Reviewed-on: https://go-review.googlesource.com/c/go/+/361918
Trust: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
3 years agoruntime: disable GC in TestPanicSystemstack's subprocess
Michael Anthony Knyszek [Fri, 5 Nov 2021 23:47:51 +0000 (23:47 +0000)]
runtime: disable GC in TestPanicSystemstack's subprocess

TestPanicSystemstack spins up a subprocess that has 2 goroutines
deadlock on a runtime lock while on the system stack, with GOMAXPROCS=2.

Each goroutine is going to be running on a P, and then is going to wedge
itself up on that P.

If a GC is active and a worker starts executing (using a P), then it
could try to preempt a goroutine that is already blocked. It won't be
able to, so it'll just sit there forever trying to suspend it. At this
point there are no more Ps to execute the remaining goroutine that needs
to print something so the parent process can continue the test.

This change fixes this issue by disabling GCs in the child process. An
alternative fix could be to increase GOMAXPROCS in the child, but maybe
letting the GC be on (which assumes it'll always be able to *eventually*
suspend a G) is just asking for trouble.

Fixes #49388.

Change-Id: I405c9dad50e24e1e68f2c52a646538da15797fbe
Reviewed-on: https://go-review.googlesource.com/c/go/+/361897
Trust: Michael Knyszek <mknyszek@google.com>
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
3 years agoruntime: disable GC during debug call tests
Michael Anthony Knyszek [Fri, 5 Nov 2021 23:35:58 +0000 (23:35 +0000)]
runtime: disable GC during debug call tests

Currently the debug call protocol implementation we use for testing is
riddled with write barriers, and called from a signal handler. This is
not safe, as write barriers need a P to execute.

Ideally this implementation would be rewritten to avoid the write
barriers, but it's not straightforward, and needs some thought. As a
temporary measure, disable GC during the debug call tests to avoid a
write barrier.

Note that this does not indicate a problem with real use of the debug
call protocol. Only our test implementation has this issue, because it
needs to get executed in a signal handler, normally a separate process
is interfacing with the protocol via process signals and ptrace (and the
like).

Fixes #49370.

Change-Id: Ic0fde5d0f4c64f9ecc9789b7dabb3954538fe0a8
Reviewed-on: https://go-review.googlesource.com/c/go/+/361896
Trust: Michael Knyszek <mknyszek@google.com>
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
3 years agocmd/compile/internal/types2: remove most asX converters (cleanup)
Robert Griesemer [Mon, 8 Nov 2021 04:08:13 +0000 (20:08 -0800)]
cmd/compile/internal/types2: remove most asX converters (cleanup)

Make it explicit in the code where we call under.
The asNamed and asTypeParam converters need to
stay: asNamed does resolution if necessary, and
asTypeParam may or may not call under() depending
on the next CL.

Reviewed uses of asNamed and .(*Named) for correctness.
Removed unnecessary Named.resolve call in lookup.

Change-Id: I2acf176925e00bd1703a00230a779aa65a8f5a51
Reviewed-on: https://go-review.googlesource.com/c/go/+/362254
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: use compiler version error when configured for compiler
Robert Griesemer [Fri, 5 Nov 2021 03:00:51 +0000 (20:00 -0700)]
cmd/compile/internal/types2: use compiler version error when configured for compiler

Fixes #49368.

Change-Id: I7c7575ae8bb6271160747e3f1888b144c3ab24c4
Reviewed-on: https://go-review.googlesource.com/c/go/+/361411
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agodoc/go1.18: strings,bytes: deprecate Title
smasher164 [Sat, 6 Nov 2021 06:59:00 +0000 (02:59 -0400)]
doc/go1.18: strings,bytes: deprecate Title

Updates #48367.

Change-Id: Ib8fc6d9dd7c3c6a70fefe077615f51a71d9c42ed
Reviewed-on: https://go-review.googlesource.com/c/go/+/361899
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Dmitri Shuralyov <dmitshur@golang.org>

3 years agonet/http: fix spelling in documentation
jiahua wang [Fri, 8 Oct 2021 01:59:15 +0000 (09:59 +0800)]
net/http: fix spelling in documentation

Change-Id: I8b0924300eafe27de98975512a78a6527a92e446
Reviewed-on: https://go-review.googlesource.com/c/go/+/354729
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Damien Neil <dneil@google.com>

3 years agoarchive/zip: don't read data descriptor early
Arran Walker [Thu, 21 Oct 2021 09:39:05 +0000 (10:39 +0100)]
archive/zip: don't read data descriptor early

Go 1.17 introduced an unnecessary change to when a zip's data descriptor
is read for file entries, how it is parsed and how the crc32 field is
used.

Before Go 1.17, the data descriptor was read immediately after a file
entry's content. This continuous read is a pattern existing applications
have come to rely upon (for example, where reads at specific offsets
might be translated to HTTP range requests).

In Go 1.17, all data descriptors are immediately read upon opening the
file. This results in scattered and non-continuous reads of the archive,
and depending on the underlying reader, might have severe performance
implications. In addition, an additional object is now initialized for
each entry, but is mostly redundant.

Previously, the crc32 field in the data descriptor would return an error
if it did not match the central directory's entry. This check has
seemingly been unintentionally removed. If the central directory crc32
is invalid and a data descriptor is present, no error is returned.

This change reverts to the previous handling of data descriptors, before
CL 312310.

Fixes #48374
Fixes #49089

Change-Id: I5df2878c4fcc9e500064e7175f3ab9727c82f100
Reviewed-on: https://go-review.googlesource.com/c/go/+/357489
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Dmitri Shuralyov <dmitshur@golang.org>

3 years agocmd/compile/internal/types2: report error for incomplete struct composite literal...
Robert Griesemer [Fri, 5 Nov 2021 04:31:08 +0000 (21:31 -0700)]
cmd/compile/internal/types2: report error for incomplete struct composite literal type

Mark a struct as "complete" with a non-nil (but possibly zero length)
fields list. Add a test when type-checking struct composite literals,
the same way we do for other composite literal types.

Fixes #49276.

Change-Id: If44a3d790bf7032ddcd155af49bdc47b1cdff4fc
Reviewed-on: https://go-review.googlesource.com/c/go/+/361412
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: clearer object string for type parameters
Robert Griesemer [Thu, 4 Nov 2021 19:50:25 +0000 (12:50 -0700)]
cmd/compile/internal/types2: clearer object string for type parameters

- print "type parameter" rather than just "type"
- print the type bound rather than the underlying type
- added an object string test

Change-Id: Ibb572ff35b74f2c6ccb27641154f096770541130
Reviewed-on: https://go-review.googlesource.com/c/go/+/361401
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: make object test an external test
Robert Griesemer [Thu, 4 Nov 2021 23:00:34 +0000 (16:00 -0700)]
cmd/compile/internal/types2: make object test an external test

Change-Id: I15c969a799404067f34f600da15b1a97d4857315
Reviewed-on: https://go-review.googlesource.com/c/go/+/361409
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: always parse tests assuming generic code
Robert Griesemer [Thu, 4 Nov 2021 23:22:49 +0000 (16:22 -0700)]
cmd/compile/internal/types2: always parse tests assuming generic code

We don't need to distinguish anymore as this is the new default.
This removes the need to prefix test package names with "generic_".

Change-Id: If9eaa0a5cffcd19deb529aca6798206f2d396ca1
Reviewed-on: https://go-review.googlesource.com/c/go/+/361408
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocrypto/x509: verification with system and custom roots
Roland Shoemaker [Fri, 1 Oct 2021 17:14:32 +0000 (10:14 -0700)]
crypto/x509: verification with system and custom roots

Make system cert pools special, such that when one has extra roots
added to it we run verifications twice, once using the platform
verifier, if available, and once using the Go verifier, merging the
results.

This change re-enables SystemCertPool on Windows, but explicitly does
not return anything from CertPool.Subjects (which matches the behavior
of macOS). CertPool.Subjects is also marked deprecated.

Fixes #46287
Fixes #16736

Change-Id: Idc1843f715ae2b2d0108e55ab942c287181a340a
Reviewed-on: https://go-review.googlesource.com/c/go/+/353589
Reviewed-by: Filippo Valsorda <filippo@golang.org>
Trust: Roland Shoemaker <roland@golang.org>

3 years agocmd/compile: avoid adding LECall to the entry block when has opendefers
hanpro [Fri, 5 Nov 2021 01:47:54 +0000 (09:47 +0800)]
cmd/compile: avoid adding LECall to the entry block when has opendefers

The openDeferRecord always insert vardef/varlive pairs into the entry block, it may destroy the mem chain when LECall's args are writing into the same block. So create a new block before that happens.

Fixes #49282

Change-Id: Ibda6c4a45d960dd412a641f5e02276f663c80785
Reviewed-on: https://go-review.googlesource.com/c/go/+/361410
Run-TryBot: Alberto Donizetti <alb.donizetti@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Alberto Donizetti <alb.donizetti@gmail.com>
Trust: Than McIntosh <thanm@google.com>
Reviewed-by: David Chase <drchase@google.com>
3 years agoall: remove more leftover // +build lines
Tobias Klauser [Sat, 6 Nov 2021 10:09:56 +0000 (11:09 +0100)]
all: remove more leftover // +build lines

CL 344955 and CL 359476 removed almost all // +build lines, but leaving
some assembly files and generating scripts. Also, some files were added
with // +build lines after CL 359476 was merged. Remove these or rename
files where more appropriate.

For #41184

Change-Id: I7eb85a498ed9788b42a636e775f261d755504ffa
Reviewed-on: https://go-review.googlesource.com/c/go/+/361480
Trust: Tobias Klauser <tobias.klauser@gmail.com>
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agocrypto/x509: use the platform verifier on iOS
Roland Shoemaker [Fri, 1 Oct 2021 16:14:10 +0000 (09:14 -0700)]
crypto/x509: use the platform verifier on iOS

Use the same certificate verification APIs on iOS as on macOS (they
share the same APIs, so we should be able to transparently use them
on both.)

Updates #46287
Fixes #38843

Change-Id: If70f99b0823dd5fa747c42ff4f20c3b625605327
Reviewed-on: https://go-review.googlesource.com/c/go/+/353403
Trust: Roland Shoemaker <roland@golang.org>
Reviewed-by: Filippo Valsorda <filippo@golang.org>
Run-TryBot: Roland Shoemaker <roland@golang.org>

3 years agoruntime: eliminate rendundant loads gcPercent
Michael Anthony Knyszek [Thu, 21 Oct 2021 21:26:50 +0000 (21:26 +0000)]
runtime: eliminate rendundant loads gcPercent

The previous change was an automated change that made gcPercent a
type-safe atomic variable. However, that introduced a lot of redundant
formal atomic loads of the variable. Remove them by only loading once in
each case, and reusing the value.

Change-Id: I49647135f423574f94506d456d1cc390150fad02
Reviewed-on: https://go-review.googlesource.com/c/go/+/357795
Trust: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
3 years agoruntime: retype gcControllerState.gcPercent as atomic.Int32
Michael Anthony Knyszek [Thu, 21 Oct 2021 21:19:23 +0000 (21:19 +0000)]
runtime: retype gcControllerState.gcPercent as atomic.Int32

[git-generate]
cd src/runtime
mv export_test.go export.go
GOROOT=$(dirname $(dirname $PWD)) rf '
  add gcControllerState.gcPercent \
// Initialized from GOGC. GOGC=off means no GC. \
gcPercent_ atomic.Int32
  ex {
    import "runtime/internal/atomic"

    var t gcControllerState
    var v, w int32
    var d int32

    t.gcPercent -> t.gcPercent_.Load()
    t.gcPercent = v -> t.gcPercent_.Store(v)
    atomic.Loadint32(&t.gcPercent) -> t.gcPercent_.Load()
    atomic.Storeint32(&t.gcPercent, v) -> t.gcPercent_.Store(v)
    atomic.Xaddint32(&t.gcPercent, d) -> t.gcPercent_.Add(d)
    atomic.Casint32(&t.gcPercent, v, w) -> t.gcPercent_.CompareAndSwap(v, w)
    atomic.Xchgint32(&t.gcPercent, v) -> t.gcPercent_.Swap(v)
  }
  rm gcControllerState.gcPercent
  mv gcControllerState.gcPercent_ gcControllerState.gcPercent
'
mv export.go export_test.go

Change-Id: I1aae34a3f782d096c6b6233bbf7986e67ce9c5f9
Reviewed-on: https://go-review.googlesource.com/c/go/+/357794
Trust: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
3 years agonet/netip: document the zero Addr
Josh Bleecher Snyder [Fri, 5 Nov 2021 21:55:52 +0000 (14:55 -0700)]
net/netip: document the zero Addr

Fixes #49364

Change-Id: I3372f80723a4deae48ef106f88b0ad880cdd0e45
Reviewed-on: https://go-review.googlesource.com/c/go/+/361914
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
3 years agodoc/go1.18: split bytes and strings packages
Dmitri Shuralyov [Fri, 5 Nov 2021 22:40:03 +0000 (18:40 -0400)]
doc/go1.18: split bytes and strings packages

A small touchup after CL 361894.

For #47694.

Change-Id: Ifc161516f897f727195d21351a3c8eda7b6e327e
Reviewed-on: https://go-review.googlesource.com/c/go/+/361895
Trust: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Carlos Amedee <carlos@golang.org>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
Run-TryBot: Carlos Amedee <carlos@golang.org>

3 years agonet/netip: add tests for Addr.AsSlice
Josh Bleecher Snyder [Fri, 5 Nov 2021 22:34:20 +0000 (15:34 -0700)]
net/netip: add tests for Addr.AsSlice

Change-Id: Ib88dd101b3bbdf4d2bfd79838994cfadef1b604d
Reviewed-on: https://go-review.googlesource.com/c/go/+/361915
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agonet/netip: add IPv4Unspecified
Jason A. Donenfeld [Wed, 3 Nov 2021 15:46:44 +0000 (16:46 +0100)]
net/netip: add IPv4Unspecified

There is IPv6Unspecified but there is not IPv4Unspecified, making for
inconsistent code. This commit adds the missing function.

Updates #49298.

Change-Id: Id2519b646323642f59fb1cc6ea8e335fdde16290
Reviewed-on: https://go-review.googlesource.com/c/go/+/361056
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Trust: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Jason A. Donenfeld <Jason@zx2c4.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
3 years agonet/netip: reduce allocations in MarshalBinary
Jason A. Donenfeld [Wed, 3 Nov 2021 14:16:01 +0000 (15:16 +0100)]
net/netip: reduce allocations in MarshalBinary

Addr's MarshalBinary required two allocations in the case of a zone
existing, and AddrPort and Prefix both required three. This refactors
things slightly so that each marshal function only needs a single
allocation.

Change-Id: I9bde9969fedc1cad64bebb607188c4287f6a0d01
Reviewed-on: https://go-review.googlesource.com/c/go/+/361054
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Run-TryBot: Jason A. Donenfeld <Jason@zx2c4.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
3 years agocrypto/x509: use platform verifier on darwin
Roland Shoemaker [Wed, 29 Sep 2021 18:31:01 +0000 (11:31 -0700)]
crypto/x509: use platform verifier on darwin

When VerifyOptions.Roots is nil, default to using the platform X.509
certificate verification APIs on darwin, rather than using the Go
verifier. Since our oldest supported version of macOS is 10.12, we are
able to use the modern verification APIs, and don't need to resort to
the complex chain building trickery employed by chromium et al.

Unfortunately there is not a clean way to programmatically add test
roots to the system trust store that the builders would tolerate. The
most obvious solution, using 'security add-trusted-cert' requires human
interaction for authorization. We could also manually add anchors to
the constructed SecTrustRef, but that would require adding a whole
bunch of plumbing for test functionality, and would mean we weren't
really testing the actual non-test path. The path I've chosen here is
to just utilize existing valid, and purposefully invalid, trusted
chains, from google.com and the badssl.com test suite. This requires
external network access, but most accurately reflects real world
contexts.

This change removes the x509.SystemCertPool() functionality, which will
be ammended in a follow-up change which supports the suggested hybrid
pool approach described in #46287.

Updates #46287
Fixes #42414
Fixes #38888
Fixes #35631
Fixes #19561

Change-Id: I17f0d6c5cb3ef8a1f2731ce3296478b28d30df46
Reviewed-on: https://go-review.googlesource.com/c/go/+/353132
Trust: Roland Shoemaker <roland@golang.org>
Run-TryBot: Roland Shoemaker <roland@golang.org>
Reviewed-by: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agonet/netip: add missing encoding.BinaryUnmarshaler to AddrPort and Prefix
Jason A. Donenfeld [Tue, 2 Nov 2021 20:33:23 +0000 (21:33 +0100)]
net/netip: add missing encoding.BinaryUnmarshaler to AddrPort and Prefix

The Addr type got an encoding.BinaryUnmarshaler implementation, but not
AddrPort and Prefix. This commit adds the missing implementation of that
interface to these types. It also adds two round trip tests that follow
the template of the existing one for Addr.

Updates #49298.

Change-Id: Iac633aed8aac579960815bb64d06ff3181214841
Reviewed-on: https://go-review.googlesource.com/c/go/+/360875
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Run-TryBot: Jason A. Donenfeld <Jason@zx2c4.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
3 years agonet/netip: add Addr.AsSlice() method
Jason A. Donenfeld [Tue, 2 Nov 2021 19:44:32 +0000 (20:44 +0100)]
net/netip: add Addr.AsSlice() method

We have AddrFrom4, AddrFrom6, AddrFromSlice and As4, As6, but we are
missing AsSlice, so this commit adds the missing function. It also gets
rid of the less ergonomic and inconsistently named IPAddrParts.

Updates #49298.

Change-Id: I1c6a2c32fc6c69b244ab49765412ffe3bbe7e5c2
Reviewed-on: https://go-review.googlesource.com/c/go/+/360874
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Jason A. Donenfeld <Jason@zx2c4.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
3 years agonet: accept "." as a valid domain name
Filippo Valsorda [Mon, 1 Nov 2021 03:22:38 +0000 (23:22 -0400)]
net: accept "." as a valid domain name

Fixes #45715

Change-Id: Ibdaa91c97d34473061b377325ebe9a3bf5696c8e
Reviewed-on: https://go-review.googlesource.com/c/go/+/360314
Trust: Filippo Valsorda <filippo@golang.org>
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agocrypto/tls: set default minimum client version to TLS 1.2
Filippo Valsorda [Mon, 1 Nov 2021 03:13:18 +0000 (23:13 -0400)]
crypto/tls: set default minimum client version to TLS 1.2

Updates #45428

Change-Id: I5d70066d4091196ec6f8bfc2edf3d78fdc0520c1
Reviewed-on: https://go-review.googlesource.com/c/go/+/359779
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Filippo Valsorda <filippo@golang.org>
Reviewed-by: Roland Shoemaker <roland@golang.org>
3 years agodoc/go1.18: fill in Go 1.18 release note TODOs using relnote
Dmitri Shuralyov [Fri, 5 Nov 2021 21:29:58 +0000 (17:29 -0400)]
doc/go1.18: fill in Go 1.18 release note TODOs using relnote

For #47694.

Change-Id: Ia80a1859bd0fc6f08d27293f519c22fd9a804fd2
Reviewed-on: https://go-review.googlesource.com/c/go/+/361894
Trust: Dmitri Shuralyov <dmitshur@golang.org>
Trust: Carlos Amedee <carlos@golang.org>
Reviewed-by: Carlos Amedee <carlos@golang.org>
3 years agocmd/go: fix bugs in replacement path handling
Michael Matloob [Tue, 19 Oct 2021 18:05:29 +0000 (14:05 -0400)]
cmd/go: fix bugs in replacement path handling

These are tested in golang.org/cl/357169 and golang.org/cl/358540.

Change-Id: I5add3f202db71731487f2688234c547abe1fd287
Reviewed-on: https://go-review.googlesource.com/c/go/+/361416
Trust: Michael Matloob <matloob@golang.org>
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agocrypto/x509: disable SHA-1 signature verification
Filippo Valsorda [Sun, 31 Oct 2021 23:23:14 +0000 (19:23 -0400)]
crypto/x509: disable SHA-1 signature verification

Updates #41682

Change-Id: Ib766d2587d54dd3aeff8ecab389741df5e8af7cc
Reviewed-on: https://go-review.googlesource.com/c/go/+/359777
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Filippo Valsorda <filippo@golang.org>
Reviewed-by: Roland Shoemaker <roland@golang.org>
3 years agonet/http: adjust a substring which matches "uninteresting" goroutine
Nevkontakte [Sat, 23 Oct 2021 14:02:25 +0000 (14:02 +0000)]
net/http: adjust a substring which matches "uninteresting" goroutine

This change allows the check to pass unmodified under GopherJS, which
stack traces are formatted differently (due to being generated by
NodeJS). There are no other functions named `interestingGoroutines` in
the standard library, so it's very unlikely to create false negatives,
and will allow reduce the number of overlays GopherJS needs to maintain.

Change-Id: I925594c6c313ca35805811f240c8528241950547
GitHub-Last-Rev: 26b32efb225c1ee85de0a51968c31b1a58d56f92
GitHub-Pull-Request: golang/go#49128
Reviewed-on: https://go-review.googlesource.com/c/go/+/358154
Reviewed-by: Damien Neil <dneil@google.com>
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
Trust: Damien Neil <dneil@google.com>
Trust: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Damien Neil <dneil@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agounicode/utf8: add AppendRune Example
jiahua wang [Thu, 7 Oct 2021 07:30:03 +0000 (15:30 +0800)]
unicode/utf8: add AppendRune Example

Also, correct TestAppendRune error message.

Change-Id: I3ca3ac7051af1ae6d449381b78efa86c2f6be8ac
Reviewed-on: https://go-review.googlesource.com/c/go/+/354529
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Trust: Robert Findley <rfindley@google.com>
Trust: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agoerrors: add errors.Unwrap example
jiahua wang [Wed, 13 Oct 2021 14:16:37 +0000 (22:16 +0800)]
errors: add errors.Unwrap example

Change-Id: Id2336a6059f7a8d627e6c0661a4d4c05485b65f3
Reviewed-on: https://go-review.googlesource.com/c/go/+/355589
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Cherry Mui <cherryyz@google.com>
Trust: Robert Findley <rfindley@google.com>

3 years agotime: fix looking for zone offset when date is close to a zone transition
zhouguangyuan [Tue, 2 Nov 2021 15:17:21 +0000 (23:17 +0800)]
time: fix looking for zone offset when date is close to a zone transition

The old implement passed start - 1 or end in func lookup to adjust the offset.But if the time is close to the last zoneTrans, like the issue, testcase and comment, the "start" from lookup will be omega. It can't be adjusted correctly.

Fixes #49284

Change-Id: Ibaf82dc4db6d5dd3279796f003d2b19c38a26341
Reviewed-on: https://go-review.googlesource.com/c/go/+/360616
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Robert Findley <rfindley@google.com>

3 years agotime: add some examples
jiahua wang [Wed, 20 Oct 2021 07:21:46 +0000 (15:21 +0800)]
time: add some examples

Change-Id: I2668cdea64f75bee87d424730d404834d69362a8
Reviewed-on: https://go-review.googlesource.com/c/go/+/357270
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Cherry Mui <cherryyz@google.com>

3 years agostrings, bytes: deprecate Title
smasher164 [Thu, 28 Oct 2021 21:08:22 +0000 (17:08 -0400)]
strings, bytes: deprecate Title

Title doesn't handle Unicode punctuation and language-specific
capitalization rules. Replace the BUG comment with a Deprecated
one, suggesting a more robust alternative, and allowing Title
to be exposed to tooling.

Fixes #48367.

Change-Id: I952f1f37cd35b587a95893fb022827bdd9ec7de9
Reviewed-on: https://go-review.googlesource.com/c/go/+/359485
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Cherry Mui <cherryyz@google.com>

3 years agodebug/plan9obj: export ErrNoSymbols
Dan Kortschak [Sun, 29 Aug 2021 04:31:03 +0000 (14:01 +0930)]
debug/plan9obj: export ErrNoSymbols

This allows callers of *File.Symbols to distinguish absence of symbols
from other errors as can already by done in debug/elf.

Fixes #48052

Change-Id: I5ba15d8473911e516c016a69c1f1c710f7fc4cd5
Reviewed-on: https://go-review.googlesource.com/c/go/+/350229
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agobytes: add example with (*Buffer).Cap, (*Buffer).Read, (*Buffer).ReadByte
jiahua wang [Tue, 12 Oct 2021 06:27:02 +0000 (14:27 +0800)]
bytes: add example with (*Buffer).Cap, (*Buffer).Read, (*Buffer).ReadByte

Change-Id: Ieb107fdfccde9f054491f667a384b16f7af71dea
Reviewed-on: https://go-review.googlesource.com/c/go/+/355289
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Cherry Mui <cherryyz@google.com>

3 years agonet/http: distinguish between timeouts and client hangups in TimeoutHandler
Charlie Getzen [Fri, 5 Nov 2021 17:27:35 +0000 (17:27 +0000)]
net/http: distinguish between timeouts and client hangups in TimeoutHandler

Fixes #48948

Change-Id: I411e3be99c7979ae289fd937388aae63d81adb59
GitHub-Last-Rev: 14abd7e4d774ed5ef63aa0a69e80fbc8b5a5af26
GitHub-Pull-Request: golang/go#48993
Reviewed-on: https://go-review.googlesource.com/c/go/+/356009
Reviewed-by: Damien Neil <dneil@google.com>
Trust: Damien Neil <dneil@google.com>
Trust: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Damien Neil <dneil@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agotime: make Ticker.Reset(0) panic
zhouguangyuan [Wed, 3 Nov 2021 14:23:29 +0000 (22:23 +0800)]
time: make Ticker.Reset(0) panic

Fixes #49315

Change-Id: I0887bad1059b25ae0749bfa1ed6ddccbecca7951
Reviewed-on: https://go-review.googlesource.com/c/go/+/361074
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
3 years agonet/http: fix comment of writeRequest
Keiichi Hirobe [Sat, 28 Aug 2021 17:38:38 +0000 (02:38 +0900)]
net/http: fix comment of writeRequest

Change-Id: I5ebfc6a89323cc086ea0e0b619370dc45da1f3a3
Reviewed-on: https://go-review.googlesource.com/c/go/+/345437
Reviewed-by: Damien Neil <dneil@google.com>
Trust: Damien Neil <dneil@google.com>
Trust: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Damien Neil <dneil@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agoruntime: fix racy stackForceMove check
Austin Clements [Fri, 5 Nov 2021 19:58:34 +0000 (15:58 -0400)]
runtime: fix racy stackForceMove check

Currently, newstack loads gp.stackguard0 twice to check for different
poison values. The race window between these two checks can lead to
unintentional stack doubling, and ultimately to stack overflows.

Specifically, newstack checks if stackguard0 is stackPreempt first,
then it checks if it's stackForceMove. If stackguard0 is set to
stackForceMove on entry, but changes to stackPreempt between the two
checks, newstack will incorrectly double the stack allocation.

Fix this by loading stackguard0 exactly once and then checking it
against different poison values.

The effect of this is relatively minor because stackForceMove is only
used by a small number of runtime tests. I found this because
mayMorestackMove uses stackForceMove aggressively, which makes this
failure mode much more likely.

Change-Id: I1f8b6a6744e45533580a3f45d7030ec2ec65a5fb
Reviewed-on: https://go-review.googlesource.com/c/go/+/361775
Trust: Austin Clements <austin@google.com>
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
3 years agocmd/go: don't assume case-insensitivity in TestScript/gcflags_patterns
Bryan C. Mills [Fri, 5 Nov 2021 19:38:57 +0000 (15:38 -0400)]
cmd/go: don't assume case-insensitivity in TestScript/gcflags_patterns

For #30432

Change-Id: I84f208705483018559b425b3669e724e7d5627ee
Reviewed-on: https://go-review.googlesource.com/c/go/+/361814
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agocrypt/aes: update formatting of ppc64le asm comments
Lynn Boger [Fri, 5 Nov 2021 17:22:50 +0000 (12:22 -0500)]
crypt/aes: update formatting of ppc64le asm comments

This does not change any code, just reformats the comments in
the asm code.

Change-Id: I70fbfa77db164898d25b59b589d3e85b8399b0fc
Reviewed-on: https://go-review.googlesource.com/c/go/+/361694
Reviewed-by: Cherry Mui <cherryyz@google.com>
Trust: Lynn Boger <laboger@linux.vnet.ibm.com>

3 years agocmd/link: enable internal linking of PIE binaries on ppc64le
Paul E. Murphy [Wed, 14 Jul 2021 19:09:27 +0000 (14:09 -0500)]
cmd/link: enable internal linking of PIE binaries on ppc64le

The amd64/arm64 relocation processing is used as a template
and updated for ppc64le.

This requires updating the TOC relocation handling code to
support linux type TOC relocations too (note, AIX uses
TOC-indirect accesses).

Noteably, the shared flag of go functions is used as a proxy
for the local entry point offset encoded in elf objects. Functions
in go ppc64le shared objects always[1] insert 2 instructions to
regenerate the TOC pointer.

[1] excepting a couple special runtime functions, see preprocess
    in obj9.go for specific details of this behavior.

Change-Id: I3646e6dc8a0a0ffe712771a976983315eae5c418
Reviewed-on: https://go-review.googlesource.com/c/go/+/352829
Run-TryBot: Paul Murphy <murp@ibm.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Lynn Boger <laboger@linux.vnet.ibm.com>

3 years agocrypto/elliptic: port P-224 and P-384 to fiat-crypto
Filippo Valsorda [Sat, 30 Oct 2021 04:27:51 +0000 (00:27 -0400)]
crypto/elliptic: port P-224 and P-384 to fiat-crypto

Also, adopt addchain code generation for field inversion, and switch
P-521 to Montgomery multiplication, which is significantly slower but
allows us to reuse the P-224/P-256/P-384 wrapper code. No one uses P-521
anyway, and it's still faster than it was in Go 1.16.

Removed a portion of tests that ran the P-224 vectors against P-256,
for some reason.

Sadly, fiat-crypto is not fast enough to replace the generic 32-bit
P-256 implementation (just yet?).

A change in visible behavior is that we literally can't internally
operate on invalid curve points anymore (yay!) but the crypto/elliptic
API locked us into accepting any pair of integers for
Add/Double/ScalarMult and return no error (sigh), although of course
that's undefined behavior. Panics are always regretted. Returning nil
leads to panics. A fixed point might be exploited. The most reasonable
solution felt to return a made up random point, which is not that
different from an off-curve point but leaks less.

name                                  old time/op    new time/op    delta
pkg:crypto/elliptic goos:darwin goarch:arm64
ScalarBaseMult/P224-8                    573µs ± 0%     146µs ± 0%   -74.56%  (p=0.000 n=7+9)
ScalarMult/P224-8                        574µs ± 0%     152µs ± 5%   -73.58%  (p=0.000 n=7+10)
MarshalUnmarshal/P224/Uncompressed-8     664ns ± 0%     481ns ± 1%   -27.64%  (p=0.000 n=8+10)
MarshalUnmarshal/P224/Compressed-8       666ns ± 1%     480ns ± 0%   -27.92%  (p=0.000 n=10+10)
pkg:crypto/ecdsa goos:darwin goarch:arm64
Sign/P224-8                              597µs ± 0%     169µs ± 2%   -71.71%  (p=0.000 n=10+9)
Verify/P224-8                           1.18ms ± 1%    0.32ms ± 5%   -72.81%  (p=0.000 n=10+10)
GenerateKey/P224-8                       577µs ± 0%     147µs ± 0%   -74.51%  (p=0.000 n=8+8)

name                                  old time/op    new time/op    delta
pkg:crypto/elliptic goos:darwin goarch:arm64
ScalarBaseMult/P384-8                   2.01ms ± 2%    0.50ms ± 0%  -75.00%  (p=0.000 n=10+8)
ScalarMult/P384-8                       2.02ms ± 3%    0.51ms ± 3%  -74.64%  (p=0.000 n=10+10)
MarshalUnmarshal/P384/Uncompressed-8    1.09µs ± 1%    0.76µs ± 0%  -30.27%  (p=0.000 n=10+9)
MarshalUnmarshal/P384/Compressed-8      1.08µs ± 0%    0.76µs ± 1%  -29.86%  (p=0.000 n=8+10)
pkg:crypto/ecdsa goos:darwin goarch:arm64
Sign/P384-8                             2.06ms ± 1%    0.56ms ± 2%  -72.76%  (p=0.000 n=10+10)
Verify/P384-8                           4.06ms ± 2%    1.08ms ± 0%  -73.49%  (p=0.000 n=10+8)
GenerateKey/P384-8                      2.01ms ± 1%    0.51ms ± 3%  -74.65%  (p=0.000 n=10+10)

name                                  old time/op    new time/op    delta
pkg:crypto/elliptic goos:darwin goarch:arm64
ScalarBaseMult/P521-8                    715µs ± 6%    1525µs ± 4%  +113.39%  (p=0.000 n=10+10)
ScalarMult/P521-8                        698µs ± 1%    1543µs ± 1%  +120.99%  (p=0.000 n=9+9)
MarshalUnmarshal/P521/Uncompressed-8     797ns ± 0%    1296ns ± 0%   +62.65%  (p=0.000 n=10+9)
MarshalUnmarshal/P521/Compressed-8       798ns ± 0%    1299ns ± 1%   +62.82%  (p=0.000 n=8+10)
pkg:crypto/ecdsa goos:darwin goarch:arm64
Sign/P521-8                              810µs ± 3%    1645µs ± 0%  +103.03%  (p=0.000 n=10+10)
Verify/P521-8                           1.42ms ± 1%    3.19ms ± 1%  +125.28%  (p=0.000 n=10+8)
GenerateKey/P521-8                       698µs ± 1%    1549µs ± 0%  +121.87%  (p=0.000 n=10+7)

Updates #40171

Change-Id: I34edf5002b5e9fad0ebb6c1e2119fb123ea6d18f
Reviewed-on: https://go-review.googlesource.com/c/go/+/360014
Run-TryBot: Filippo Valsorda <filippo@golang.org>
Trust: Filippo Valsorda <filippo@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Julie Qiu <julie@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agoruntime: skip TestNoShrinkStackWhileParking on netbsd/arm64
Bryan C. Mills [Fri, 5 Nov 2021 14:11:11 +0000 (10:11 -0400)]
runtime: skip TestNoShrinkStackWhileParking on netbsd/arm64

This test appears to deadlock frequently on the only netbsd-arm64
builder we have (netbsd-arm64-bsiegert). Skip the test to provide
more useful test coverage for other failures.

For #49382

Change-Id: I3be32f58ce1e396f7c69163e70cf58f779f57ac6
Reviewed-on: https://go-review.googlesource.com/c/go/+/361615
Trust: Bryan C. Mills <bcmills@google.com>
Trust: Benny Siegert <bsiegert@gmail.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Benny Siegert <bsiegert@gmail.com>
3 years agoruntime: make consistentHeapStats acquire/release nosplit
Michael Anthony Knyszek [Wed, 3 Nov 2021 18:49:51 +0000 (18:49 +0000)]
runtime: make consistentHeapStats acquire/release nosplit

consistentHeapStats is updated during a stack allocation, so a stack
growth during an acquire or release could cause another acquire to
happen before the operation completes fully. This may lead to an invalid
sequence number.

Fixes #49395.

Change-Id: I41ce3393dff80201793e053d4d6394d7b211a5b7
Reviewed-on: https://go-review.googlesource.com/c/go/+/361158
Trust: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Austin Clements <austin@google.com>
3 years agoruntime: pace the scavenger according to 1% of overall CPU time
Michael Anthony Knyszek [Mon, 25 Oct 2021 21:49:05 +0000 (21:49 +0000)]
runtime: pace the scavenger according to 1% of overall CPU time

Currently the scavenger is paced to 1% of 1 CPU because it had
scalability problems. As of the last few CLs, that should be largely
resolved. This change resolves the TODO and paces the scavenger
according to 1% of overall CPU time.

This change is made separately to allow it to be more easily rolled
back.

Change-Id: I1ab4de24ba41c564960701634a128a813c55ece9
Reviewed-on: https://go-review.googlesource.com/c/go/+/358675
Trust: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
3 years agoruntime: use a controller to control the scavenge rate
Michael Anthony Knyszek [Mon, 4 Oct 2021 21:44:06 +0000 (21:44 +0000)]
runtime: use a controller to control the scavenge rate

Currently the scavenge rate is determined by a bunch of ad-hoc
mechanisms. Just use a controller instead, now that we have one.

To facilitate this, the scavenger now attempts to scavenge for at least
1 ms at a time, because any less and the timer system is too imprecise to
give useful feedback to the controller. Also increase the amount that we
scavenge at once, to try to reduce the overheads involved (at the
expense of a little bit of latency).

This change also modifies the controller to accept an update period,
because it's useful to allow that to be variable.

Change-Id: I8a15b2355d0a7c6cbac68c957082d5819618f7d7
Reviewed-on: https://go-review.googlesource.com/c/go/+/353975
Trust: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
3 years agoruntime: add harddecommit GODEBUG flag
Michael Anthony Knyszek [Mon, 18 Oct 2021 19:01:33 +0000 (19:01 +0000)]
runtime: add harddecommit GODEBUG flag

This change adds a new debug flag that makes the runtime map pages
PROT_NONE in sysUnused on Linux, in addition to the usual madvise calls.
This behavior mimics the behavior of decommit on Windows, and is helpful
in debugging the scavenger. sysUsed is also updated to re-map the pages
as PROT_READ|PROT_WRITE, mimicing Windows' explicit commit behavior.

Change-Id: Iaac5fcd0e6920bd1d0e753dd4e7f0c0b128fe842
Reviewed-on: https://go-review.googlesource.com/c/go/+/356612
Trust: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
3 years agoruntime: don't hold the heap lock while scavenging
Michael Anthony Knyszek [Mon, 4 Oct 2021 20:36:49 +0000 (20:36 +0000)]
runtime: don't hold the heap lock while scavenging

This change modifies the scavenger to no longer hold the heap lock while
actively scavenging pages. To achieve this, the change also:
* Reverses the locking behavior of the (*pageAlloc).scavenge API, to
  only acquire the heap lock when necessary.
* Introduces a new lock on the scavenger-related fields in a pageAlloc
  so that access to those fields doesn't require the heap lock. There
  are a few places in the scavenge path, notably reservation, that
  requires synchronization. The heap lock is far too heavy handed for
  this case.
* Changes the scavenger to marks pages that are actively being scavenged
  as allocated, and "frees" them back to the page allocator the usual
  way.
* Lifts the heap-growth scavenging code out of mheap.grow, where the
  heap lock is held, and into allocSpan, just after the lock is
  released. Releasing the lock during mheap.grow is not feasible if we
  want to ensure that allocation always makes progress (post-growth,
  another allocator could come in and take all that space, forcing the
  goroutine that just grew the heap to do so again).

This change means that the scavenger now must do more work for each
scavenge, but it is also now much more scalable. Although in theory it's
not great by always taking the locked paths in the page allocator, it
takes advantage of some properties of the allocator:
* Most of the time, the scavenger will be working with one page at a
  time. The page allocator's locked path is optimized for this case.
* On the allocation path, it doesn't need to do the find operation at
  all; it can go straight to setting bits for the range and updating the
  summary structure.

Change-Id: Ie941d5e7c05dcc96476795c63fef74bcafc2a0f1
Reviewed-on: https://go-review.googlesource.com/c/go/+/353974
Trust: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
3 years agosyscall: fix mkall.sh for openbsd/386
Tobias Klauser [Fri, 5 Nov 2021 16:09:22 +0000 (17:09 +0100)]
syscall: fix mkall.sh for openbsd/386

CL 287654 converted the syscall package on openbsd/386 to use libc.
However, the mksyscall.pl invocation wasn't adjusted. Do so now to use
syscall_openbsd_libc.go like the other libc-based openbsd ports.

Change-Id: I48a7bd6ce4c25eca5222f560ed584e412b466111
Reviewed-on: https://go-review.googlesource.com/c/go/+/361481
Trust: Tobias Klauser <tobias.klauser@gmail.com>
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agonet/netip: optimize As4 and As16
Josh Bleecher Snyder [Fri, 5 Nov 2021 17:07:11 +0000 (10:07 -0700)]
net/netip: optimize As4 and As16

name    old time/op  new time/op  delta
As16-8  2.88ns ± 3%  2.16ns ± 3%  -25.19%  (p=0.000 n=15+15)

Fixes #49379
Updates #20859

Change-Id: If4cf58d19ed0e2ac0f179da5c132ed37061e4cb7
Reviewed-on: https://go-review.googlesource.com/c/go/+/361674
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
3 years agoruntime: fix a lock rank ordering and some edges.
Dan Scales [Thu, 4 Nov 2021 23:03:49 +0000 (16:03 -0700)]
runtime: fix a lock rank ordering and some edges.

The first stack-trace in #49361 shows that traceBuf must precede fin in
lockrank ordering, since traceBuf is acquired in StartTrace(), which
eventually leads to getting fin in queueFinalizer(). It is fine to move
traceBuf above fin, since there are no other conflicting dependencies.

The second stack trace shows that there is an edge bewtween reflectOffs
and fin, since reflectOffs is acquired in addReflectOff, and map
operations can lead to an allocation that eventually causes fin to be
acquired in queueFinalizer().

Fixes #49361

Change-Id: I8e857ef9ecdff37fdd229e4dba22e15bc71d4ba5
Reviewed-on: https://go-review.googlesource.com/c/go/+/361407
Trust: Dan Scales <danscales@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
3 years agonet/netip: don't accept ParseAddr with leading zeros
Brad Fitzpatrick [Fri, 5 Nov 2021 04:35:55 +0000 (21:35 -0700)]
net/netip: don't accept ParseAddr with leading zeros

Fixes #49365
Updates #30999

Change-Id: Ic92bce01b435baf70574c65524bde82f9cee3d8d
Reviewed-on: https://go-review.googlesource.com/c/go/+/361534
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Trust: Brad Fitzpatrick <bradfitz@golang.org>

3 years agoruntime: mark TestTracebackArgs test functions nosplit
Cherry Mui [Thu, 4 Nov 2021 20:53:30 +0000 (16:53 -0400)]
runtime: mark TestTracebackArgs test functions nosplit

The argument liveness tests expect outputs where a dead stack slot
has a poisoned value. If the test function is preempted at the
prologue, it will go with the morestack code path which will spill
all the argument registers. Mark them nosplit to avoid that.

Should fix #49354.

Change-Id: I3b13e72e925748687a53c494bfaa70f07d9496fa
Reviewed-on: https://go-review.googlesource.com/c/go/+/361211
Trust: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
3 years agocmd/compile: only update source type when processing struct/array
Cuong Manh Le [Fri, 5 Nov 2021 13:30:38 +0000 (20:30 +0700)]
cmd/compile: only update source type when processing struct/array

CL 360057 fixed missing update source type in storeArgOrLoad. However,
we should only update the type when processing struct/array. If we
update the type right before calling storeArgOrLoad, we may generate a
value with invalid type, e.g, OpStructSelect with non-struct type.

Fixes #49378

Change-Id: Ib7e10f72f818880f550aae5c9f653db463ce29b0
Reviewed-on: https://go-review.googlesource.com/c/go/+/361594
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
3 years agonet: skip new alloc test on some builders
Brad Fitzpatrick [Fri, 5 Nov 2021 15:18:57 +0000 (08:18 -0700)]
net: skip new alloc test on some builders

Change-Id: Ib785212eb9e035a10d327696c180caf730d2e5e7
Reviewed-on: https://go-review.googlesource.com/c/go/+/361634
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Trust: Brad Fitzpatrick <bradfitz@golang.org>

3 years agocmd/compile: gracefully fallback when inline bodies are missing
Matthew Dempsky [Thu, 4 Nov 2021 20:28:25 +0000 (13:28 -0700)]
cmd/compile: gracefully fallback when inline bodies are missing

Currently, we rely on a "crawling" step during export to identify
function and method bodies that need to be exported or re-exported so
we can trim out unnecessary ones and reduce build artifact sizes. To
catch cases where we expect a function to be inlinable but we failed
to export its body, we made this condition a fatal compiler error.

However, with generics, it's much harder to perfectly identify all
function bodies that need to be exported; and several attempts at
tweaking the algorithm have resulted in still having failure cases.
So for now, this CL changes a missing inline body into a graceful
failure instead.

Change-Id: I04b0872d0dcaae9c3de473e92ce584e4ec6fd782
Reviewed-on: https://go-review.googlesource.com/c/go/+/361403
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Dan Scales <danscales@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Dan Scales <danscales@google.com>
3 years agointernal/poll: fix the wrong function names in comments
Andy Pan [Fri, 5 Nov 2021 02:40:31 +0000 (10:40 +0800)]
internal/poll: fix the wrong function names in comments

Change-Id: Ib17df2751209129ad6d1c148829625b347b702a9
Reviewed-on: https://go-review.googlesource.com/c/go/+/361514
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Brad Fitzpatrick <bradfitz@golang.org>

3 years agobufio: fix typo in comment
lgbgbl [Fri, 5 Nov 2021 04:57:59 +0000 (04:57 +0000)]
bufio: fix typo in comment

Change-Id: I6074334bc73ee1773fc83d049142ad764e39e5a9
GitHub-Last-Rev: 6a8789318ab132a169a89be807394af55c855ce9
GitHub-Pull-Request: golang/go#49373
Reviewed-on: https://go-review.googlesource.com/c/go/+/361413
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agocrypto/ecdsa: draw a fixed amount of entropy while signing
Filippo Valsorda [Mon, 27 Sep 2021 18:40:06 +0000 (14:40 -0400)]
crypto/ecdsa: draw a fixed amount of entropy while signing

The current code, introduced in CL 2422, mixes K bits of entropy with
the private key and message digest to generate the signature nonce,
where K is half the bit size of the curve. While the ECDLP complexity
(and hence security level) of a curve is half its bit size, the birthday
bound on K bits is only K/2. For P-224, this means we should expect a
collision after 2^56 signatures over the same message with the same key.

A collision, which is unlikely, would still not be a major practical
concern, because the scheme would fall back to a secure deterministic
signature scheme, and simply leak the fact that the two signed messages
are the same (which is presumably already public).

Still, we can simplify the code and remove the eventuality by always
drawing 256 bits of entropy.

Change-Id: I58097bd3cfc9283503e38751c924c53d271af92b
Reviewed-on: https://go-review.googlesource.com/c/go/+/352530
Trust: Filippo Valsorda <filippo@golang.org>
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Roland Shoemaker <roland@golang.org>
3 years agoruntime: add a maymorestack hook that moves the stack
Austin Clements [Mon, 27 Sep 2021 14:05:13 +0000 (10:05 -0400)]
runtime: add a maymorestack hook that moves the stack

This adds a maymorestack hook that moves the stack at every
cooperative preemption point.

For #48297.

Change-Id: Ic15f9bcbc163345e6422586302d57fda4744caec
Reviewed-on: https://go-review.googlesource.com/c/go/+/359797
Trust: Austin Clements <austin@google.com>
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: David Chase <drchase@google.com>
3 years agoruntime: add always-preempt maymorestack hook
Austin Clements [Wed, 15 Sep 2021 21:14:21 +0000 (17:14 -0400)]
runtime: add always-preempt maymorestack hook

This adds a maymorestack hook that forces a preemption at every
possible cooperative preemption point. This would have helped us catch
several recent preemption-related bugs earlier, including #47302,
 #47304, and #47441.

For #48297.

Change-Id: Ib82c973589c8a7223900e1842913b8591938fb9f
Reviewed-on: https://go-review.googlesource.com/c/go/+/359796
Trust: Austin Clements <austin@google.com>
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: David Chase <drchase@google.com>
3 years agocmd/{asm,compile,internal/obj}: add "maymorestack" support
Austin Clements [Tue, 20 Aug 2019 21:39:09 +0000 (17:39 -0400)]
cmd/{asm,compile,internal/obj}: add "maymorestack" support

This adds a debugging hook for optionally calling a "maymorestack"
function in the prologue of any function that might call morestack
(whether it does at run time or not). The maymorestack function will
let us improve lock checking and add debugging modes that stress
function preemption and stack growth.

Passes toolstash-check -all (except on js/wasm, where toolstash
appears to be broken)

Fixes #48297.

Change-Id: I27197947482b329af75dafb9971fc0d3a52eaf31
Reviewed-on: https://go-review.googlesource.com/c/go/+/359795
Trust: Austin Clements <austin@google.com>
Run-TryBot: Austin Clements <austin@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agocmd/compile,cmd/internal/objabi: move -d flag parser to objabi
Austin Clements [Fri, 17 Sep 2021 14:07:41 +0000 (10:07 -0400)]
cmd/compile,cmd/internal/objabi: move -d flag parser to objabi

This moves and slightly generalizes the -d debug flag parser from
cmd/compile/internal/base to cmd/internal/objabi so that we can use
the same debug flag syntax in other tools.

This makes a few minor tweaks to implementation details. The flag
itself is now just a flag.Value that gets constructed explicitly,
rather than at init time, and we've cleaned up the implementation a
little (e.g., using a map instead of a linear search of a slice). The
help text is now automatically alphabetized. Rather than describing
the values of some flags in the help text footer, we simply include it
in the flags' help text and make sure multi-line help text renders
sensibly.

For #48297.

Change-Id: Id373ee3b767e456be483fb28c110d025149be532
Reviewed-on: https://go-review.googlesource.com/c/go/+/359956
Trust: Austin Clements <austin@google.com>
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>